By TOM ABATE
San Francisco Chronicle
December 16, 2008
After all, it took nearly a year for economists to decide that a recession had started last December.
Even those trying to improve the predictive power of economics make light of their track record.
"I regularly hear the accusation that economic forecasting is no better than weather forecasting, but this does a disservice to weather forecasters," joked Jon Faust, a former Federal Reserve Board economist who is now a professor of at Johns Hopkins University.
"When one does economic forecasting, you have to realize your forecast is going to be wrong," said Harvard economist James Stock.
"That said, forecasting is an important challenge. You can't just throw up your hands and quit," he said.
Stock and his Princeton University collaborator, Mark Watson, have helped lay the theoretical foundation for a new generation of computer models that try to forecast where the economy is going by getting a clearer picture of where it stands today.
One of the most prominent of these new forecasting systems is published on a monthly basis by the Chicago Federal Reserve. It is called the Chicago Fed National Activity Index.
Scott Brave, the Chicago Fed economist who is developing the system, said the index captures 85 economic indicators, including inflation, unemployment, manufacturing output, retail sales and housing activity. Its computer model boils down all 85 variables into one number that takes the temperature of the economy, in a manner of speaking.
Zero is the baseline temperature. Brave said a reading of zero indicates an economy growing at just the right balance between inflation and recession.
If the synthesis of all 85 variables spits out a positive number, that suggests the economy is heating up and could be headed for inflation. Conversely, a string of negative readings would suggest that the economy is cooling down and heading toward recession.
Brave said a three-month average of the index seems to give the best reading by smoothing out any month-to-month variations between hot and cold.
Right now this three-month average is well below zero, as would be expected during a recession. In fact, the average dipped into the danger zone in December 2007 -- which happens to be when the panel of economists, charged with dating recessions by looking backward at the data, decided that this downturn began.
But Brave is careful not to describe the index as a predictive tool. "It's a useful summary measure of a lot of data," he said.
Amy Resnick, editor in chief of Bond Buyer, publishes the index in her daily financial newspaper but doesn't swear by it.
"We publish a lot of data," she said, putting her decision to add the Chicago Fed's index of indexes in the "more is better category."
Stock, the Harvard theorist, said one reason economic forecasting is harder than weather forecasting is that people can and do change the rules.
"You're trying to predict a system with people in it who keep doing things that make the problem more complicated," he said.
Charles Macal, a computer-modeling scientist at Argonne National Laboratory in Illinois, is working on that next level of challenge -- trying to predict the behavior of large groups of people inside economic systems.
His computerized model tries to take into account how factors like age, gender, income, education and employment affect economic behavior.
"One way to think about it is: What do we have to specify about the state of the individuals that is relevant to what they buy and sell," Macal said.
The work, which is still in its embryonic stages, is being carried out on a computer cluster that Macal has nicknamed Cassandra. It's a reference to the character from Greek mythology who was blessed with the ability to see the future -- but cursed by not being able to convince people she was right.
"That's the modeler's lament," Macal said.
Distributed to subscribers for publication by
Scripps Howard News Service, http://www.scrippsnews.com
Publish A Letter in SitNews Read Letters/Opinions